SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Atoms is building Physical AI—real-world robots for industries including food, mining, and transport. This Safety Operations Lead role owns the operational safety discipline at scale, transforming safety from a documentation exercise into a core operating system that runs at the speed of the fleet.
You will design and operate the end-to-end incident management pipeline: detection, notification, severity classification, on-scene coordination, root cause investigation, corrective action tracking, and closure. You'll build the event triage system that converts raw field telemetry, disengagements, operator reports, and near-miss data into a ranked, actionable queue with clear taxonomy and severity models. You'll lead investigations into collisions, disengagements, and anomalous behavior, driving them to defensible causal accounts and tracking corrective actions to verified closure.
Additionally, you'll design and own the training and recurrent-qualification program for safety drivers, autonomous vehicle operators, and remote support staff, including curriculum, practical evaluation, and qualification records. You'll establish operational safety metrics, reporting cadence, and ensure internal records withstand external scrutiny. You'll be the critical feedback channel between field operations and engineering, ensuring operational evidence feeds hazard analysis and product change assessment.
This is a hands-on role. You will write most processes yourself, personally run the first dozen investigations, and only then hire to scale. The work is AI-native: you'll use LLM-based tooling to query and summarize event data, draft investigation reports, classify event backlogs, and identify patterns across incidents. You'll build small tools rather than wait on roadmaps, while maintaining the boundary that AI output is a hypothesis and drafting accelerant—every safety conclusion requires a human who has examined the evidence and will stand behind it.
Success milestones: 45 days—incident response documented and exercised, event taxonomy in place, clear assessment of weakest areas. 3 months—triage running predictably with published metrics, investigations producing tracked reports, training program rebuilt with documented records, AI tooling identified. 12 months—system runs autonomously, tooling implemented and in daily use, next hire identified.
